Unfortunately the Lamnidis et al. 2018 did not test the origin of the Siberian ancestry in present-day Finns and Russians. I reveal it now. Ancient samples found from Bolshoy gives high F3-results for both groups. I included five Finnish groups, three of my own sample selection and two smaller selections from two academic sources.
FinnMostCW - least drifted Finnish group (- outliers)
FinnLocal - most drifted Finnish group (- outliers)
Finn21M - most Scandinavian group (- outliers)
Both academic Finnish groups look incoherent. I wonder why.
It is obvious that the Levaluhta admixture added to the present-day European backbone in Finns (negative F3-values) doesn't explain all non-Europeanness and there have to be an unknown third component.
Source1, source2, target, f_3, std.err, Z and SNP count
Additionally, here are resultw of my old admixture analysis based on Dna.Land's program. Although it sounds a weird idea to make admixture analysis of ancient samples using present-day populations, it is a normal practice in many academic studies. Studies usually use Nganasans to represent Siberians together with ancient samples. It is reasonable to say that the gene flow can be detected also backwards. It is also very likely that the gene flow between contemporary populations has been bidirectional. Nevertheless, I am not fully convinced about the gene flow direction regarding the European side of Finns and Northern Russians. F3-statistics gave several European candidates and the big picture is more complex.
Levaluhta (JK excluding JK2065 outlier)
Finnic 43,7
Saami 30,0
Uralic 12,1
Siberian 9,5
Baltic 2,2
East_Asian 1,9
Bolshoy
Uralic 38.9
Siberian 29.5
Saami 11.6
Finnic 7.1
East_Asian 6.1
Northeast_European 5.4
Baltic 1.5
It's mostly the fault of the f3 method, more extreme sources always produce higher f3 scores even if they are not optimal. Sweden and Bolshoy or Lithuanian and Bolshoy will be lower than Sardinian and Ulchi.
ReplyDeleteCan be, but it is also obvious that the Finns can't be modelled by two ancient populations. I am also bound to use same methods with geneticists, doing something else wouldn't make sense.
DeleteIf you use qpadm, you can try more than two. Bolshoy is enough to explain the Siberian.
DeleteThe reason why Nganasans give better fit is that
Delete- European + Nganasan
gives for Finns a better F3 fit than
- European + any population consisting of Finns + Nganasans.
The additional Finnish admixture in another population makes the fit worse, which is against common sense, but it is what it should be. If the Levaluhta already includes Finnish admixture, it makes it a poor fit. Such a fit can be poorer than Sardinian + Ulchi.
Sardinian Ulchi FinnMostCW1M -0.002540 0.000545 -4.659 870927
Sardinian+nganasan gives a better fit than balts/swedes+nganasans anyway. It's just that f3 is attracted to extremes instead of the real source populations.
DeleteI can do qpAdm, but it wouldn't prevent the dilemma I already mentioned (ancient populations that are already have Finnish admixture). The Bolshoy can be a better fit than Levaluhta, but it doesn't mean that it is the right one. More wrong conclusions made by readers don't make the result better.
DeleteF3 is not attracted to extremes, it only weights PURE drifts without giving importance to real admixture sizes. That is why Lithuanian + Nganasan is a better fit than Lithuanian + Karelian (or Mari or Levaluhta). The same problem exists with qpAdm.
Sardinian+ulchi or nganasan wouldn't work with qpAdm and in f3 would look better than a fit that would work with qpadm. F3 doesn't try to model real sources.
ReplyDeleteModern Finns didn't even exist until late Iron Age so Bolshoy does not have Finnish admixture.
But qpAdm still works with pure drifts and admixture proportions are based on it, not on real admixtures.
DeleteI wrote " It is also very likely that the gene flow between contemporary populations has been bidirectional. ". So the Finnish admixture in Bolshoy doesn't mean Finnish, but some ancient component derived from that ancient component to Finns. Of course it was not Finnish thousands years ago and they didn't speak Finnish. Similarly the Finns have many ancient admixtures/components which are not present in reference populations we just use in our admixture analysis and are therefor understood only Finnish.
Yes but it's about finding the source populations closest to reality. Nothing more eastern than Bolshoy came to Finland later than Bolshoy, quite the opposite in fact as a few people have seen a sneak peek of dozens of samples.
DeleteI have seen qpAdm with Nganasans in Finland.
DeleteDoesn't matter, you have yourself done a qpadm showing She in Finland and the UK but neither Nganasans or She were really in those places. The populations that came to Finland after Bolshoy were all more western than Bolshoy.
DeletePalaan vielä tuohon Anthrogenicassa julkaistuun sekoitusajoon (https://anthrogenica.com/showthread.php?15941-First-ancient-DNA-from-mainland-Finland-reveals-origins-of-Siberian-ancestry-in-regio/page11).
ReplyDeleteValitettavasti erityisesti levikset ovat huonolaatuisia, joten niiden perusteella ei ilmeisesti voi tehdä tarkkoja vertailuja, mutta olisiko mitenkään mahdollista nykyisillä näytteillä pystyä erottamaan erityisesti Chalmny-Varreen liittyvää paleo-skandinaavista perimää ja länsiuralilaista perimää nykysaamelaisissa? Tuo ajo antoi viitteitä siitä, että länsiuralilaisissa oli ylimääräistä EHG (Sidelkino) ja Latvian metsästäjä -perimää, jota löytyy sekä suomalaisista että saamelaisista. Tämän suuntaisia tietojahan on ymmärtäkseni tihkunut myös kohta julkaistavista näytteistä. Saisiko tätä perimää siis jotenkin selvennettyä ilman noita kyseisen ajon puutteita?
Ehkä olen vain kärsimätön ja meidän pitäisi vain odottaa, että näköpiirissä olevat näytteet julkaistaan.
I have not redistered in Anthrogenica and do not see those PCA's.
DeleteThe first question is who were Western Uralic people in relation to recently living people. We can test them, see the EHG admixture and then test Finns, Saamis and Estonians, see the difference between ancient samples and them. If the EHG admixture reveals something we certainly have something, but I am not very convinced about that it happens. We likely see soon ancient N1c1 samples from Estonia or Latvia holding EHG and say "they were Western Uralic people". How we know it if we cannot place then into the continuum of Finns and Estonians.
Speaking about qpAdm and admixtures. QpAdm is a powerful tool, but only if all source population are isolations, ancients like EHG and WHG or present-day like Nganasans, but it fails if we try to use admixed populations like Levaluhta or Saami as sources. A+B is fine in mapping genetic drifts, but A+(A+B) does not work. We have another choice, we can use Admixture. It can solve A+(A+B),but has another weakness. It fails to map correctly those isolations.
For the sake of clarity, there is two kinds of Admixture runs, unsupervised snd supervised. Isolations can be run in a controlled manner by using the supervised mode, like they usually do in studies. The bad thing is that doing so we force all tested populations into supervised admixtures, even if they don't have those admixtures.
DeleteA+AB does work in qpadm, like modeling Mordvas as Polish+western Scythian worked in your own test even though these share ancestry.
DeleteI meant the case Polish + (Polish + smthg). Doudle drift worsens the fit. Adm statistics doesn't tell everything. Better drift fits are achieved by isolations. It is easy to test:
DeleteFinnLocal FinnMostCW Finnx 0.000285 0.000277 1.030 906484
If you have f.ex. EHG in outgroups and excessive EHG in test groups you get faulty results.
The Scythian you used is as Polish as Levanluhta is Finnish :)
DeleteUnohdin kiittää tästä vastauksesta! Se menee joka tapauksessa asian ytimeen! :)
ReplyDeleteKristiina, kirjoita mieluummin englanniksi. Lukijoista puolet ei ymmärrä suomea ja englannin käyttö saattaa toimia esimoderointina.
DeleteKristiina,
Deleteyou wrote on another blog that there are infinite number of models. This is not true, there are only many time frames. Are you trolling? I don't care what people write on internet, but this stood out. I posted my answer there, but the blog keeper set the moderator mode on and is not active now. Here is my answer:
"There is an infinite amount of possible modellings. To get a more balanced view, you should take into account M. Myllylä's post: http://terheninenmaa.blogspot.com/2018/12/finnish-and-north-russian-ancestries.html#comment-form"
We have two point of views here, in one we can explain more recent populations by using EHG, WHG, CHG, ENA and early farmers. This helps us to map a large number of populations over Eurasian in a way we can compare results, but of cource that kind of results don't explain the whole history. Already Bronze Age populations were more mixed and present-day populations are mixtures of Bronze Age and Iron Age populations. In a best practice we should use contemporary popelations as admixtures, but for more practical reason it is not possible. Anyway, if the time frame between used admixture populations is big we have a risk to fail, because same admixtures are hidden.
If you think I am trolling, then it is true, and I need not post any unwanted comments. It is a pity, because I believe in the plurality of opinions and freedom of speech. If my opinions are not tollerated, I am very happy not to express them.
ReplyDeleteI have seen so many nMonte models from amateurs and qpAdm models from Academics that I doubt that a model put by someone is the ultimate truth. Even in the academic papers, models are often contradictory between themselves: e.g. in one model there is EMBA, in the other MLBA. One paper says that there is no Steppe while others adamantly argue that there is Steppe. We have now hundreads of ancient samples and we can use most of them for nMonte analyses. If we let the program choose freely, we get very many sources, i.e. model is overfitted. If we choose the right sources, we force the model to comply with our preferred view of the history. If you put in the analysis a more western source, it is compensated by a more eastern source to get a similar result. On top of this, you can even model East Asians using Western Eurasians and you will get a result. There are certainly good practices, but it seems permissible to use a contemporary source from the same burial as a source for the others. In addition, you yourself said that the program gives a better rating for a model which uses isolated populations which are archaeologically a less probable source. You can use Baikal Glazkovo_EBA to model Leväluhta people and you will get a result, but the mtDNA and yDNA of Glazkovo people are completely different and the model is not archaeologically plausible. In the Uralic case, the most important thing to keep in mind is that we are not using samples from the most plausible cultures. It would be great to have samples from Serteja, Lyalovo, Valdaj, Volsosovo, Finnish Comb Ceramic, Koškino Ceramic, West Siberian Comb Ceramic, Pečora-Dvina Comb Ceramic, Swedish TRB, Finnish Corded Ware, Fatyanovo, Abashevo, Garino Bor, Textile Ceramics, Imitated Ceramics, Kiukainen Culture, Krotovo, etc. We could use all of them in our modelings. There is an endless list of archaelogical cultures that could be used in different combinations in the analyses if we only had DNA samples.
You wrote that there is infinite number of models. Of course we can use 28 letters to combine infinite numbers of nonsense. You are right and I am wrong. I also have experience about intolerant people and closed-mindedness from the nearest past, you are not the only one :)
DeleteMost of your observations are right, but please notice that everything regarding the history of people and nations is extreme difficult to determine and we have to be patient. For example, F3-statistics can't give unambiguous truths, but it CAN tell what is impossible. We know that Nganasans never invaded Finland, but we can't deny that similar in Finland we have similar genes. We can test what is possible. To know how all those cultures you mentioned are connected to us, we ought to have alot contemporary samples from all time frames. It probably will never happen. It is much easier to use very old isolations, like WHG, EHG etc., or use modern samples and at first look who are closest us and then try to go backwards in history. Picking rousing cultures and samples does the task difficulter.